Interactive visualization with R - Network Graphs - 2

One should look for what is and not what he thinks should be. (Albert Einstein)

Module completion checklist

Objective Complete
Create nodes and edges dataframes
Build and customize a network HTMLwidget

Creating the network: edges

  • Now that the dataset is prepped for visualization, we must generate the edge dataframe and the node dataframe that will comprise our network graph

  • We can start by transforming our similarity matrix into an edge dataframe

  • The edge dataframe informs visNetwork:

    • to draw an edge from which node to which node
    • the value (the thickness) of the edge
  • We can do this using the tidy function from the broom package, which turns the messy output of built-in R functions into tidy dataframes
library(broom)
# Create edge dataframe. 
hds_edges = tidy(hds_sim)
# Edges dataframe has to be named this way for visNetwork input.
colnames(hds_edges) = c("from", "to", "value")
head(hds_edges)
# A tibble: 6 x 3
  from  to    value
  <fct> <fct> <dbl>
1 1     3     0.455
2 1     4     0.734
3 1     5     0.747
4 1     6     0.800
5 1     7     0.299
6 1     8     0.405

Setting a similarity threshold

  • We need to choose a similarity threshold to create edges for the network
  • A similarity threshold is a measure of the strength of the relationship between nodes in the network
  • The wrong similarity threshold will result in a very sparse network graph
    • A sparse network graph is a type of network graph in which only a fraction of all possible connections between nodes (vertices) exists.
  • The threshold value can be chosen via trial and error based on what works best for your network
  • We generally assign the threshold as 0.5 or as the mean/median of the similarity matrix

Setting the number of edges

  • To simplify our network viz, we are going to subset the first 200 edges, ranked according to their values
# We choose the median as the threshold since this gives us the best visualization.
hds_edges = subset(hds_edges,value>median(hds_edges$value))

# Arrange by order of edge thickness.
hds_edges = arrange(hds_edges, desc(value))

# Subset only top 200.
hds_edges = hds_edges[1:200,]

Creating the network: nodes

  • Now that we have the edges set, we can focus on the nodes
  • The nodes input to visNetwork must have an id column
  • We can get these nodes by extracting the unique nodes from the from and to columns of the edges dataframe
# Get unique nodes from edges dataframe and combine them 
hds_nodes_from = data.frame(id = unique(hds_edges$from))
hds_nodes_to = data.frame(id = unique(hds_edges$to))
hds_nodes = rbind(hds_nodes_from,hds_nodes_to)

# Retain unique nodes in case nodes are repeated in `from` and `to` columns
hds_nodes = unique(hds_nodes)

Creating the network: nodes

  • It can also have additional attributes like color, shape, labels, etc.
    • Color : Used to visually distinguish nodes based on categories, attributes, or characteristics.
    • Shape : Determines the geometric form of nodes, emphasizing groupings or types within the network.
    • Label : Determines the geometric form of nodes, emphasizing groupings or types within the network.
# Add color to the nodes dataframe based on stroke value from original dataframe
hds_small = select(hds_small, stroke)               #<- we only need the target info
hds_small$id = rownames(hds_small)                 

# Merge nodes dataframe with the dataframe with Target value 
hds_nodes = merge(hds_nodes, hds_small,
                    by = "id", all.x = TRUE)            #<- merge() needs the `id` column to 
                                                        #   join the two dataframes
# Assign color to nodes based on the stroke value
hds_nodes$color = factor(hds_nodes$stroke,          #<- create a factor 
labels = c("orange", "darkblue"), #<- assign color 
levels = c(1, 0))                                 #   based on Target value

# Because we aligned colors based on the stroke column, we can drop it
hds_nodes = select(hds_nodes, c(id, color))

Creating the network: nodes (cont’d)

head(hds_nodes)
    id    color
1 1002 darkblue
2 1019 darkblue
3 1035 darkblue
4 1036 darkblue
5 1038 darkblue
6 1044 darkblue

Module completion checklist

Objective Complete
Create nodes and edges dataframes

✔

Build and customize a network HTMLwidget

Creating the network

  • The nodes and edges dataframes are all we need to create the visNetwork
  • Let’s take a look at the resulting network
  • You can zoom in to see the individual nodes
# Create network.
hds_network = visNetwork(hds_nodes,                 #<- set nodes
                           hds_edges)           #<- set edges
hds_network

centered

Analyzing the network

  • Share your thoughts and findings in the chat:

    • Which target has the most nodes?
    • Are the nodes for vulnerable people connected to the nodes for non-vulnerable people?
    • What could this tell us about the similarity between the attributes of people in these two categories?

centered

Analyzing the network

  • We notice that there are:
    • more nodes with stroke 0 (non-vulnerable people)
    • fewer nodes with stroke = 1 (vulnerable people)
  • The nodes for vulnerable people do not appear very connected to those for non-vulnerable people
  • This means that the attributes of people from different categories are not very similar

centered

Customizing the network visualization

  • Since we built the network graph as a widget, we can make it more interactive to incorporate additional information

  • For example, we can add extra functionality using visOptions, such as:

    • Highlighting the nearest nodes when a single node is selected
    • Creating a dropdown menu to select specific nodes 
# Add network visualizations
hds_network = visNetwork(hds_nodes,                 #<- set nodes
                           hds_edges)  %>%            #<- set edges
                 visOptions(highlightNearest = TRUE,    #<- highlight nearest nodes
                                                        #   when clicking on a node 
                           nodesIdSelection = TRUE)     #<- create a dropdown menu to 
                                                        #   select particular nodes

Customizing the network visualization (cont’d)

hds_network

centered

Saving networks with htmlwidgets

  • It is possible to save the networks as HTML files, allowing us to share them, use them later, or embed them in presentations
# Set working directory to where you save interactive plots.
setwd(plot_dir)

# Load the library.
library(htmlwidgets)

# Save desired interactive plot to an HTML file.
saveWidget(hds_network,                     #<- plot object to save
           "network.html",                    #<- name of file to where the plot is to be saved
            selfcontained = TRUE)             #<- set `selfcontained` to TRUE, so that 
                                              #   all necessary files and scripts are embedded
                                              #   into the HTML file itself

Knowledge check

centered

Link: Click here to complete the knowledge check

Exercise

centered


You are now ready to try tasks 3-7 in the Exercise for this topic

Module completion checklist

Objective Complete
Create nodes and edges dataframes

✔

Build and customize a network HTMLwidget

✔

Network graphs: topic summary

In this part of the course, we have covered:

  • Transforming and preparing data for a network graph visualization
  • Building and customizing interactive network graphs

Congratulations on completing this module!

icon-left-bottom